The Builder's Journal

What I'm building. What I'm learning.

March 2026

Giving AI a Memory: Publishing Engram and Redesigning the Product Page

We published Engram to GitHub, hardened the repo, then redesigned the /engram page from 3 generic sections to 7 content-rich sections — porting the README's strongest arguments directly onto the web.

Read →

Day in the Build: CRO Polish, BIOS Pipeline, and the Middleware Bug That Nearly Froze Our Payouts

The workshop10 min read

A full build day: /start page CRO with 14-product comparison table, a visual BIOS pipeline replacing dead text, Phase 5 debug library, and a middleware fix that unblocked Stripe verification with 3 days to spare.

Read →

AriaOS: Tenant Zero and the Future of Multi-Agent Orchestration

The story12 min read

AriaOS started as an internal tool for managing our own agent teams. Now it's evolving into a multi-tenant platform. Here's the vision — and here's what we're building first.

Read →

Why Context-First Beats Prompt Engineering

The system8 min read

Prompt engineering optimizes the question. Context-First architecture optimizes the intelligence. Here's why that distinction changes everything.

Read →

The Flywheel: How the System Gets Smarter Over Time

The system12 min read

The Context-First methodology isn't a 12-step process with a finish line. Step 12 feeds back into Step 3 — creating a flywheel where every cycle makes the system more intelligent, more accurate, and more valuable.

Read →

The Agentic Workflow System: 18 Commands That Build Anything

The system14 min read

Every project follows the same discipline — plan, research, discuss, execute, verify. Our 18-workflow system is the operating manual that turns methodology into repeatable execution.

Read →

From BIOS to Production: Deploying Agents to Real Platforms

The workshop12 min read

An agent that lives in a chat window isn't an agent that generates revenue. Here's how BIOS-emerged agents get deployed to Shopify, Klaviyo, Meta, and Stripe — where the real work happens.

Read →

Agentic Orchestration: Running Multi-Agent Teams at Scale

The workshop14 min read

One agent is useful. Five agents working together are transformative. But coordination doesn't happen by accident — it requires orchestration, routing, and a kernel that keeps everyone aligned.

Read →

February 2026

How I Cut Ad Spend 78% and Profits Went Up

The proof10 min readceltic-knot

Celtic Knot was spending $377K per quarter on ads. After installing a BIOS-powered agent team, we cut to $83K — and revenue improved. Here's exactly how.

Read →

eComX Build Log — Phase 4: Content as Infrastructure

The workshop12 min read

The content layer was the hardest phase — not because of complexity, but because content systems require different thinking. MDX pipelines, SEO infrastructure, case studies, and this journal. 45 commits to make content a first-class citizen.

Read →

eComX Build Log — Phase 3: 47 Commits to a Payment Engine

The workshop14 min read

The conversion layer: Stripe payments, Resend email sequences, cookie-based auth, cron-driven drip campaigns, full-funnel analytics, and CSRF protection. 47 commits, 8 plans, zero payment failures.

Read →

The Autonomy Ladder: Observer to Sovereign

The system12 min read

Not every agent should have the same permissions. The 5-tier autonomy ladder defines how much freedom an agent earns — from observing human work to running operations independently.

Read →

Atomic Planning: How We Ship 121 Commits in 4 Days

The workshop14 min read

Building eComX from zero to production-ready in 4 working days required extreme planning discipline. Here's the atomic planning methodology — phase by phase, plan by plan, task by task.

Read →

eComX Build Log — Phase 1: Foundation in 28 Commits

The workshop14 min read

Day one of building eComX: Next.js 15, Supabase, design tokens, the debug dashboard, and every decision made along the way. A complete build log from empty directory to functional foundation.

Read →

Agent Libraries: Teaching AI to Remember and Get Sharper

The workshop14 min read

After agents emerge and pass validation, they build their own operational infrastructure — 1,600-line playbooks, restraint doctrines, intelligence threads, and context-efficient indexes. This is how agents go from capable to excellent.

Read →

Building a Multi-Tenant AI Platform from Zero

The workshop9 min readariaos

AriaOS is the Context-First methodology as a product. Here's the technical deep dive — governance engine, autonomy tiers, orchestrator loop, and why we're building it.

Read →

Cross-Platform Validation: When 4 AIs Check Each Other's Work

The system14 min read

One AI can hallucinate. Two can agree on the same hallucination. But when four LLMs independently review the same artifact — blindly, iteratively, until convergence — you get something closer to truth.

Read →

The Multi-Model Synthesis: What Happens When 4 AIs Disagree

The story12 min read

ChatGPT, Claude, Gemini, and Grok each see the world differently. When you give them the same strategic problem and compare their answers, the disagreements are where the real insights hide.

Read →

How I Discovered That AI Agents Need to Be Born, Not Built

The story7 min read

I tried designing AI agents from scratch. They were useless. Then I let them emerge from real customer data — and everything changed.

Read →

How the AXIS Scientific Team Emerged

The story12 min read

The same methodology that sells Irish jewelry also co-authored a published physics hypothesis. Here's how four AI agents formed a scientific research team using the Agentic Emergence process.

Read →

Agent Emergence: How AI Agents Are Born From Data

The story15 min read

Most people design AI agents top-down — pick a name, assign a role, write a prompt. We do the opposite. Our agents emerge from data, self-name, and define their own capabilities. Here's the 10-step process.

Read →

7 Projects, 5 Domains, One Year

The future6 min read

A look at what's next — AriaOS launch, KohWork Songkran release, Emmalish completion, and the convergence of everything I've been building.

Read →

Your BIOS Is Alive: Versioning Brand Intelligence

The system10 min read

The BIOS isn't a deliverable you file away. It's a living system that evolves with your brand — versioned, tracked, and continuously refined as new data changes what your AI agents know.

Read →

SYS Loader: Export Your BIOS to Any LLM Project

The system10 min read

Your BIOS shouldn't be locked into one tool. SYS Loader packages 33 specs into portable documents that any AI platform can consume — Claude, ChatGPT, Gemini, or your own custom system.

Read →

Full-Funnel Measurement: The Dual-Send Architecture

The proof12 min read

Client-side pixels miss 30% of conversions. Server-side alone misses behavioral context. The dual-send architecture fires both — and achieves an EMQ score of 8.0+ that directly improves ad delivery.

Read →

BIOS: 33 Specs From Real Data, Not Templates

The system15 min read

The Brand Intelligence Operating System isn't a brand guidelines document. It's 33 machine-readable specifications generated from your data warehouse — the constraint architecture that makes AI agents genuinely intelligent about your brand.

Read →

January 2026

907 Customer Comments to 5 Brand Agents: The Data Pipeline

The proof14 min read

Celtic Knot had 6 years of customer data sitting unused in Shopify, Klaviyo, and spreadsheets. Here's how we turned 907 product reviews into the raw material that birthed 5 AI agents who actually understand the brand.

Read →

The Data Warehouse: Every Signal Your Brand Has Ever Made

The system14 min read

Before AI can understand your brand, it needs memory. The data warehouse captures every purchase, every comment, every review, every email — every surface your brand has ever touched.

Read →

Shopify OS 2.0 Governance: Deterministic Build Systems for Commerce

The workshop12 min read

Building on Shopify without governance is chaos. Sections break. Styles conflict. Schema limits silently fail. Here's the governance framework that makes Shopify themes deterministic and reliable.

Read →

The Integration Layer: APIs, CLIs, and MCP

The system14 min read

Before the data warehouse. Before the BIOS. Before agents. There's the integration layer — the nervous system that connects every platform your brand touches.

Read →

Klaviyo Performance: Email as a Revenue Engine

The proof12 min read

Most brands treat email as a notification channel. Celtic Knot's BIOS-driven agents treat it as a revenue engine — with segment-specific voice modulation, heritage-anchored sequences, and measurable revenue per subscriber.

Read →

Setting Up: From Zero to Building in 30 Minutes

The system12 min read

Every project starts the same way — not with code, but with environment. Here's how we go from empty laptop to production-ready workspace in under 30 minutes.

Read →

Get the journal weekly.

New entries, methodology insights, and build updates delivered every Thursday.